Hacker News Reader

I love it, just purchased a pack. I've also found that it is a very great tool to test LLM, like take a screenshot of a half resolved game and feed it to ChatGPT with the rules and ask him to select the next target

View on HN

Thank you so much! Also, you might find this interesting regarding testing LLMs: https://www.nicksypteras.com/blog/cbs-benchmark.html

turn out Claude Sonnet 4.5 is far better as resolving those as ChatGPT 5.2

Reply on news.ycombinator.com